The problem. A truth set is only as trustworthy as the assumptions behind it. If everyone validates variant callers against one reference (say, GIAB), systematic blind spots in that reference become invisible — the whole field agrees on the same wrong answers. Independent benchmarks, built differently, are how you check.

The idea. Platinum Genomes derives a high-confidence variant set for NA12878 using a fundamentally different principle: pedigree consistency. By sequencing a family and requiring variants to segregate according to Mendelian inheritance, it flags calls that violate the pedigree as suspect — yielding millions of phased, validated variants with an orthogonal notion of “confident.”

Why it matters. For validating variant_calling_nf, having Platinum alongside GIAB means I can check a caller against two independently constructed answer keys. Where they agree, I’m confident; where they differ, there’s something to understand about the benchmark, not just the caller. Reading it reinforces a habit that generalises well beyond variants: never trust a single gold standard.

Verdict. Foundational for rigorous variant-calling evaluation, and a neat demonstration of using inheritance as a validation signal — the same Mendelian logic from day 5, put to work. Read it paired with GIAB as the two-benchmark discipline.